effects of delayed ettringite formation on reinforced

HAL Id: hal-02155138https://hal.insa-toulouse.fr/hal-02155138

Submitted on 13 Jun 2019

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Effects of delayed ettringite formation on reinforcedconcrete structures

Yvan Thiebaut, Stéphane Multon, Alain Sellier, Laurie Lacarriere, LaurentBoutillon, Djemal Belili, Lionel Linger, François Cussigh, Sofiane Hadji

To cite this version:Yvan Thiebaut, Stéphane Multon, Alain Sellier, Laurie Lacarriere, Laurent Boutillon, et al.. Effects ofdelayed ettringite formation on reinforced concrete structures. 12th fib International PhD Symposiumin Civil Engineering, 2018, Prague, Czech Republic. pp.1193-1200. �hal-02155138�

https://hal.insa-toulouse.fr/hal-02155138

https://hal.archives-ouvertes.fr

Proc. of the 12th fib International PhD Symposium in Civil Engineering

Aug 29 to 31, 2018, Czech Technical University in Prague, Prague, Czech Republic 1

Effects of delayed ettringite formation on reinforced concrete structures

Yvan Thiebaut a,b , Stéphane Multon a, Alain Sellier a, Laurie Lacarrière a, Lau-

rent Boutillon b, Djemal Belili c, Lionel Linger b, François Cussigh d, Sofiane

Hadji e

a Université de Toulouse, LMDC (Laboratoire Matériaux et Durabilité des Constructions),

UPS/INSA Génie Civil, 135 avenue de Rangueil, 31077 Toulouse Cedex 04, France

b Vinci Construction Grands Projets, Direction Scientifique, 5, cours Ferdinand-de-Lesseps

92 851 Rueil-Malmaison Cedex, France

c Cofiroute, 12-14 rue Louis Blériot. 92500 Rueil-Malmaison Cedex, France

d Vinci Construction France, 61 avenue Jules Quentin 92730 Nanterre, France

e Sixence Concrete, 24 rue Jean-Baptiste Huet 78350 Jouy-en-Josas, France

Abstract The prediction and reassessment of mechanical behaviour of reinforced structures affected by delayed

ettringite formation (DEF) is a major challenge for structure managers. Firstly, several experimental

tests were performed in laboratory to study the influence of both uni and tri-axial reinforcements on

DEF expansion. Strain decreases in reinforced directions were observed, proving that DEF expansion

under restraint is anisotropic. Cracks were observed parallel to the restrained directions. No strain trans-

fert occurs from restrained directions to other ones. Secondly, data provided by these results are used

to fit a numerical finite element model taking into account both chemical and mechanical aspects of

DEF. Finally, an element of a DEF damaged structure is modelled with the whole model and compared

to on site observations.

1 Introduction

Delayed ettringite formation (DEF) is a pathology that can occur in cementicious material under certain

circumstances, resulting in expansion and damage [1]–[4]. In the context of reinforced structures, ex-

pansions decrease in restrained or loaded directions [2]–[6]. Reassessment and prediction of the me-

chanical behaviour of such structures are necessary. To do so, both non linear reinforced concrete be-

haviour and DEF impacts have to be modelled [7], [8]. The finite element model proposed by Sellier et

al. takes into account creep and drying strains, localised and diffuse damages and internal pressures [9],

[10]. Anisotropic behaviour of DEF expansions under anisotropic stresses is considered through inde-

pendent plasticity criteria in each principal direction of stress. Before structural application, investiga-

tions were conducted to quantify the reinforcements and stresses influence on expansive concrete. Some

authors studied effects of homogeneous restraint imposed by external reinforcements on immersed

mortar prisms (40×40×160 mm) affected by DEF [3]. In other studies, reinforced concrete beams

(610×914×5486 mm and 250×500×3000 mm) in heterogeneous hydric and mechanical conditions were

considered [4], [6], [11]. In all these studies, expansions decreased in restrained directions. Comple-

mentary experimental study was necessary to extend those existing results to reinforced “pathologic”

concrete. Non-reinforced, uni-axially and tri-axially reinforced prisms (100×100×500 mm) were ex-

posed to a thermal cycle reaching 80°C and then stored in heated and non-renewed water, in homoge-

nous thermal, mechanical and saturation conditions. Measurements of alkali concentration evolution in

the storage bath due to leaching were made along the immersion period. Strains were regularly meas-

ured on the concrete surface with a calibrated extensometer. Materials, experimental set-up and condi-

tions are presented in the first part of this article. Focus is then made on the model used in this study,

especially on DEF mechanical modelling. The last section is dedicated to experimental results analysis,

specimens modelling and reassessement of a reinforced concrete beam affected by DEF.

12th fib International PhD Symposium in Civil Engineering

2 Durability and life assessment

2 Experimental investigation on reinforced concrete

2.1 Materials, curing conditions and storage

Experiments were performed on concrete with 410 kg/m3 of standard Ordinary Portland Cement CEM

I 52.5 R (Table1 - left) and a water/cement ratio of 0.48. Non-reactive regarding Alkali-Silica Reaction

(NF P 18-590) siliceous aggregates were used, supplied from Palvadeau quarry (France). Cement com-

position was obtained by Inductively Coupled Plasma - Optical Emission Spectrometry (ICP-OES) and

chromatography analysis (Table 1 - right). Sodium hydroxide, NaOH, was added to the mixing water

at a content of 5.0 kg/m3 of concrete.

Table 1 Concrete composition (left) and cement chemical composition (right).

Component Proportion (kg/m3) Oxides Chromatography and ICP-OES analysis

Cement 410 SiO2 20.1

Water 197 Al2O3 4.52

Sand 0/0.315 93.5 Fe2O3 2.37

Sand 0.315/1 174 CaO 64.6

Sand 1/2 184 MgO 0.78

Sand 2/4 196 SO3 3.19

Gravel 4/8 188 K2O 0.16

Gravel 8/12.5 878 Na2O* 0.15 (before NaOH addition)

NaOH addition 5.0 Na2Oeq 0.25 (before NaOH addition)

Fig. 1 Non-reinforced (a.), axially reinforced (b.) and triaxially reinforced (c.) specimens

(100×100×500 mm). Strain measurement orientation on concrete surface (d.).

A 14 days long thermal autogeneous treatment has been applied to the specimens after casting. The

temperature increased from 23 to 80°C in 33 hours, decreased slowly from 80 to 72°C in 63 hours and

returned from 72 to 23°C in 240 hours (Fig. 1 - e). The specimens were then stored at 23°C and 50%

of relative humidity during 14 additionnal days. After that period, they were immersed in agitated, non-

renewed water maintained at 38°C. The concrete/water volume ratio of the storage bath was constant

d.

e.


Yvan Thiebaut, Stéphane Multon, Alain Sellier, Laurie Lacarrière, Laurent Boutillon, Djemal Belili, Lionel Linger, François Cussigh, Sofiane Hadji

3

and equal to 0.23. Complementary physicochemical investigations, which are not detailed here in order

to focus on mechanical aspects, have been made to provide data for numerical modelling.

2.2 Experimental set-up and measurements

Specimens were 100×100×500 mm prisms. Stress-free expansions were studied on plain prismatic

specimens (Fig. 1 - a). In the case of DEF expansion under uni-axial restraint (Fig. 1 - b), a HA12

central longitudinal reinforcement was added in order to reach a section ratio between steel and concrete

equal to 1.14%. Steel plates and restraint systems were added in order to improve longitudinal bars

anchoring. Finally, tri-axially restrained specimens were composed of a HA12 central longitudinal re-

inforcement and external transversal stirrups distributed along the specimens (Fig. 1 - c). The transver-

sal reinforcement ratio was equal to 0.68%. An extensometer was used for strain measurements. Stain-

less steel studs were glued on each face of the specimens. As the specimen width was not sufficient to

allow direct measurements of transversal strains, some studs were glued forming angles of 45° and 135°

with the longitudinal direction (Fig. 1 - d). Regular strain measurements were performed during the 333

days of specimen immersion.

2.3 Characterization

Tensile tests were performed on HA12 steel bars used as longitudinal reinforcements and their Young’s

modulus, yield strength and ultimate strength were 220 GPa, 505 MPa and 609 MPa, respectively.

Table 2 presents the concrete Young’s modulus, compressive and tensile strength. A minimum of three

tests for each were performed on cylindrical specimens, submitted or not to heat treatment. A decrease

of modulus and strength was observed, probably due on one hand to the impact of heat treatment on

the microstructure development during hydration and on the other hand to differential thermal expan-

sion between paste and aggregates, damaging the Intefacial Transition Zone.

Table 2 Heated (80°C) and non-heated (20°C) concrete characterization 28 days after casting.

Specimens designation E (MPa) Rc (MPa) Rt (MPa)

20 °C-28d 40170 ± 550 40.0 ± 1.6 3.9 ± 0.4

80 °C-28d 35420 ± 270 31.3 ± 0.9 3.0 ± 0.3

3 DEF modelling principle

3.1 General mechanical modelling principle

Fig. 2 Idealized rheologic scheme for DEF poromechanical model

When precipitating in the finite volume of porosity, delayed ettringite crystals act on pore edges. This

internal pressure provoques matrix microcracking, leading to expansion. Compressive stresses in con-

crete (induced by loading or restraint due to reinforcement) can prevent microcracking apparition and

propagation. The model used is dedicated to numerical simulation of non linear reinforced concrete

behaviour taking into account strains due to creep and drying, localised and diffuse damages and inter-

nal pressures, including the one induced by DEF. A poromechanical framework is considered to man-

age interactions between phenomena: expansive products formed in pores acts on the solid matrix as



an internal pressure, which leads to creep and damage in tension, as schemed in Fig. 2. Reversible creep

modelling relies on a Kelvin solid rheologic module when permanent creep is modelled thanks to a

Maxwell visco-elastic fluid rheologic module. Water effects on concrete strains, such as drying creep,

are modelled through a poromechanical capillary pressure theory: tension between liquid/gas water and

solid matrix are assimilated to intraporal pressure. Numerous plasticity criteria are used to assess plastic

strains. Localised cracking in tension and crack reclosure are managed by three Rankine criteria, one

on each principal directions. Three sets of three others Rankine criteria are used to model diffuse matrix

cracking due to alkali-silica reaction, DEF and water capillary pressure. Finally, failure in shear/com-

pression is modelled through a Drucker Prager criteria with non associated flow. Damage comes along

with each plastic phenomena described above, with creep and with temperature. The main aspects of

DEF mechanical modelling are broached below.

3.2 DEF internal pressure

A poromechanical law is used to relate the intraporal pressure PDEF with the difference between the

volume of delayed ettringite DEF, assessed from a chemical model, and the volume available for de-

layed ettringite precipitation (1). The volume avalilable for delayed ettringite precipitation is composed

of a fraction of the initial accessible porosity v, DEF, of the volume created by poral strains under pres-

sure and external load (tr()-tr(pl)), and of the volume created by DEF diffuse microcracking

(tr(plDEF)). The Biot coefficient approximation bDEF is obtained thanks to a homogenization method.

The link between the difference of volumes and the intraporal pressure is made by the Biot’s modulus

MDEF, assessed from the Biot coefficient and the compressibility moduli of the delayed ettringite and

of the matrix deprived of the volume occupied by the delayed ettringite.

(1)

3.3 DEF plasticity criteria and swelling anisotropy under restraint

If the intraporal pressure exceeds the solid skeleton strength in the case of an unloaded material, diffuse

inelastic strains appear, corresponding to microcracking (plDEF in Fig. 2 and eq. (1)). To numerically

do so, three plastic Rankine criteria applied on principal stresses directions are used (2). The maximal

stress value due to DEF intraporal pressure in the solid skeleton strength of the material is obtained

mutiplying this pressure PDEF by a stress concentration factor C, and is compared to the effective tensile

strength �̃�𝐼𝑡. In the case of a loaded material, only negative total undamaged stresses ⟨�̃�𝐼 ∈ [𝐼,𝐼𝐼,𝐼𝐼𝐼]⟩

− (i.e.

compression) are taken into account. In presence of anisotropic compressive loading, the plastic swell-

ing associated to these cracks becomes anisotropic because the criteria have different values (Fig. 3).

(2)

Fig. 3 Anisotropic cracking induced by an anisotropic compressive loading.



5

3.4 Distributed reinforcements

Model application to reinforced concrete structures could be complicated by the precision needed to

mesh those reinforcements. A solution involves considering their structural effects as equivalent to the

ones due to orthotropic distributed reinforcements. This method, avoiding explicit meshing of rein-

forcements, has been used in this article. Different reinforcements can be considered simultaneously

with different orientations if necessary. Reinforcement plasticity and relaxation were modelled [12].

Perfect bond between matrix and reinforcements were considered in the cases of application presented

below.

4 Analysis and modelling

4.1 Experimental and numerical results

For each case of reinforcement, experimental and numerical expansions were plotted on the same figure

(Fig. 4). First, the focus is made on the analysis of the experimental results (cross symbol). Final longi-

tudinal and transversal strains of 0.59% and 0.66%, respectively, were reached in the case of non-

reinforced concrete (Fig. 4 - a), confirming observations made by Bouzabata et al. [3] concerning iso-

tropic behaviour of DEF expansion in stress-free conditions. Effects of reinforcements on expansion

are visible comparing the longitudinal strains of non-reinforced and uni-axially reinforced concrete

(Fig. 4 - b). In this last case, reinforced direction exibits strains lower than 0.24% (decrease of about

60%). Cracks were observed parallel to the restrained directions. The longitudinal strain decrease is

similar for the tri-axially reinforced concrete specimens (Fig. 4 - c). Stainless steel stirrups limited

greatly transversal expansions, from 0.66% in stress-free conditions to 0.33% when restrained. Conse-

quently, DEF expansion under anisotropic restraint is anisotropic. Moreover, uni-axially reinforced

specimens exibit transversal strains slightly lower than ones of non-reinforced specimens. Longitudinal

restraint has no major impact on transversal expansion: no strain transfer occurs. Comparison between

longitudinal expansions of uni-axially, equal to 0.24%, and tri-axially reinforced specimens, equal to

0.21% seems to confirm that each direction could be considered independent from the restraint applied

on the others. Nevertheless, some authors [3] observed on longitudinally restrained specimens that

transversal expansions were slightly larger than for stress-free specimens, but without sufficient gap to

consider a complete transfer of DEF expansions from loaded to free directions. Bouzabata et al.’s ex-

perimental program was made on highly reactive mortar samples (free expansions reached 2.3%): both

the differences of material used in each study and their reactivity could explain these distinct phenom-

ena.

Numerical results were obtained using several input data provided by the cement composition, con-

crete formulation and characterization, curing and storage conditions, and reinforcement characteris-

tics. Alkali content in the intraporal water of the specimens was assessed with a diffusion model. The

main parameters of alkali diffusion in concrete were fitted on experimental measurements of leached

alkali in the storage water. Delayed ettringite volume evolution was provided by a chemical model

using sulfate and aluminium initial contents, temperature, porosity saturation and alkali evolution as

input data [13]. Then, the mechanical model described in chapter 3 has been used to simulate DEF

effects on concrete specimens. In the case of structure requalification, expansion tests on non-reinforced

concrete extracted from reinforced structures are mostly made years after casting and provide infor-

mations on residual expansion only. In this context, the most reliable data available for model calibra-

tion and validation are the strains and cracks distribution, opening and orientation, measured directly

on the damaged structures. So, it appeared here more relevant to calibrate the model mainly on experi-

mental expansions of reinforced specimens. Stress-free expansions are slightly underestimated (Fig. 4

- a): after 361 days of immersion, numerical strains are equal to 0.51% versus 0.59% and 0.66% for

experimental longitudinal and transversal strains, respectively. Expansion kinetics is consistent regard-

ing to experimental results. No clear numerical distinction occurs between longitudinal and transversal

strains. Both longitudinal and transversal expansions of uni-axially reinforced specimens are well-as-

sessed (Fig. 4 - b): in the reinforced direction, numerical and experimental strains are respectively

0.21% and 0.24% and are 0.57% and 0.58% in stress-free transversal direction. Concerning tri-axially

reinforced specimens, a parametric study was made on stainless steel characteristics because no tensile

test has been performed on it. Young’s modulus value was supposed equal to 200 000 MPa. Two cases

are presented here. On the one hand, yield strength and strain hardening coefficient were equal 160



MPa and 52 000 MPa, respectively. On the other hand, yield strength and strain hardening coefficient

were equal 600 MPa and 2 000 MPa, respectively. Whatever the case considered, both longitudinal and

transversal strains remain slightly overestimated (Fig 4 - c). When low yield strength and high strain

hardening coefficient are used, longitudinal strains are well assessed, equal to 0.23% numerically and

0.21% experimentally, but the gap between modelling and measurements is important in transversal

direction with values of 0.46% and 0.33%, respectively. This difference is reduced considering the case

of high yield strength and low strain hardening coefficient: assessed transversal strains are equal to

0.38%. Longitudinal expansion provided by modelling, equal to 0.26%, remains consistent with exper-

imental ones.

Fig. 4 Effect of restraint on expansion: experimental and numerical results

4.2 Application to reinforced concrete beam

The case of an isostatic reinforced concrete beam affected by DEF has been analysed. The beam length

between posts is 9.2 m, with a transversal section of 1.0 m height and 0.7 m width (Fig. 5). Longitudinal

reinforcement ratio is around 2%. In transversal direction, this ratio is equal to 0.35% in the first 1.6 m

close to the post, and is only 0.17% in the central span. Their localisation was taken into account thanks

to 0.1 m thick distributed orthotropic reinforcement zones on each edge of the beam. Two symetries

has been used to decrease the calculation time. Homogeneous chemical expansion was considered,

meaning that heat treatment, water content and alkali content were supposed homogeneous. Distributed

vertical loading of 0.07 MN/m was applied on the upper surface of the beam. Posts have been modelled

as punctual supports. Beam deformed shape before DEF beginning appears realistic (Fig. 6 - left). Cur-

vature is positive (bent concave upward) because of the distributed loading applied on the upper part of

the structure. Lower reinforcements are in tension, when upper ones are in compression (Fig. 7 - left).

Localised macrocracks appeared in the lower part of the beam mid-span, because of tensile stress in

concrete (Fig. 7 - right). Vertical displacement of the lower axis is close to 0.005 m.

Incertainty over

transversal restraint



7

Fig. 5 Beam geometry and reinforcements. Equivalent distributed reinforcement ratio considered

in the model.

Fig. 6 Beam deformed shapes (amplification equal to 50) before (left) and after (right) DEF reac-

tion.

Fig. 7 Stress evolution in reinforcements (left) and concrete (right) in lower and upper axis of

beam mid-span.

During DEF reaction, concrete expansion conducted to an increase of tensile stress in reinforcements

resulting in compression in concrete. This phenomenon occurs mainly in heavily reinforced areas, such

as the beam lower part (Fig. 7 - blue curves). After 140 days of expansion, compressive stresses lead to

tensile macrocracks reclosure. Stress evolution in concrete lower part appears slightly unstable: it could

be due to stress values projection from the Gauss points, where they are assessed, to the mesh, and to

convergence issues induced by competition between DEF diffuse microcracks propagation and mac-

rocracks reclosure. Because of asymmetric vertical distribution of longitudinal reinforcements in the

beam, expansions are more important in upper than in lower area resulting in gradual beam curvature

decrease (Fig. 6 - right). Initial vertical maximal displacement (caused by the mechanical loading) de-

creases from 0.005 m to less than 0.002 m in 350 days. Numerical results are consistent with on site

observations: in some case, beam curvature can even change, from positive value before DEF beginning

to negative value years later.

-100

0

100

200

300

400

500

600

0 100 200 300 400

σS

(MP

a)

t (days)

lower axis

upper axis

-10,0

-8,0

-6,0

-4,0

-2,0

0,0

2,0

4,0

0 100 200 300 400

σC

(MP

a)

t (days)

lower axis

upper axis



5 Conclusion

Interaction between reinforcements and expansions occuring in concrete affected by DEF has been

studied in the present work. Experimental expansions of stress-free, uni-axially and tri-axially concrete

specimens were studied and compared. Several important issues for the development of structural mod-

elling have been highlighted. Firsly, expansion induced by DEF is isotropic in stress-free conditions.

Secondly, experimental evidence of anisotropic concrete swelling due to DEF under anisotropic loading

has been made. Cracks were mostly parallel to the restrained directions. And finally, no strain transfert

occurs from restrained directions to others, meaning that each direction could be considered independ-

ent from the restraint applied on the others. The second part of this article has been dedicated to numer-

ical modelling of these specimens. Concrete behaviour such as creep, plasticity and damage were con-

sidered along with DEF reactions and mechanical consequences. Numerical results appeared realistic

compared to experimentation. Structural application has been made on a reinforced concrete beam af-

fected by DEF. The consequences of interactions between reinforcements and expansions on structural

mechanical behaviour appeared consistent with on site observations.

Acknowledgements

This work took part in a collaboration between LMDC laboratory (Laboratoire Matériaux et Durabilité

des Constructions) based in Toulouse and “VINCI Construction Grands Projets” Scientific Department,

“VINCI Construction France”, “Cofiroute” and “Sixense Concrete”.

References

[1] Z. Zhang, J. Olek, and S. Diamond, ‘Studies on delayed ettringite formation in early-age, heat-

cured mortars: I. Expansion measurements, changes in dynamic modulus of elasticity, and

weight gains’, Cem. Concr. Res., vol. 32, no. 11, pp. 1729–1736, Nov. 2002.

[2] X. Brunetaud, L. Divet, and D. Damidot, ‘Impact of unrestrained Delayed Ettringite Formation-

induced expansion on concrete mechanical properties’, Cem. Concr. Res., vol. 38, no. 11, pp.

1343–1348, Nov. 2008.

[3] H. Bouzabata, S. Multon, A. Sellier, and H. Houari, ‘Effects of restraint on expansion due to

delayed ettringite formation’, Cem. Concr. Res., vol. 42, no. 7, pp. 1024–1031, Jul. 2012.

[4] M. M. Karthik, J. B. Mander, and S. Hurlebaus, ‘Deterioration data of a large-scale reinforced

concrete specimen with severe ASR/DEF deterioration’, Constr. Build. Mater., vol. 124, no.

Supplement C, pp. 20–30, Oct. 2016.

[5] B. Burgher, A. Thibonnier, K. J. Folliard, T. Ley, and M. Thomas, ‘Investigation of the internal

stresses causes by delayed ettringite formation in concrete’, Center for Transportation Research,

the University of Texas at Austin, Technical report 0–5218, 2008.

[6] D. J. Deschenes, O. Bayrak, and K. J. Folliard, ‘ASR/DEF-damaged bent caps : shear tests and

field implications’, Ferguson Structural Engineering Laboratory - University of Texas at Austin,

Austin, Texas, Technical report 12–8XXIA006, Aug. 2009.

[7] N. Baghdadi, F. Toutlemonde, and J.-F. Seignol, ‘Modélisation de l’effet des contraintes sur

l’anisotropie de l’expansion dans les bétons atteints de réactions de gonflement interne’, in Con-

férence AUGC, Bordeaux, 2007.

[8] E. Grimal, A. Sellier, S. Multon, Y. Le Pape, and E. Bourdarot, ‘Concrete modelling for exper-

tise of structures affected by alkali aggregate reaction’, Cem. Concr. Res., vol. 40, no. 4, pp.

502–507, Apr. 2010.

[9] A. Sellier, S. Multon, L. Buffo-Lacarrière, T. Vidal, X. Bourbon, and G. Camps, ‘Concrete

creep modelling for structural applications: non-linearity, multi-axiality, hydration, temperature

and drying effects’, Cem. Concr. Res., vol. 79, pp. 301–315, Jan. 2016.

[10] P. Morenon, S. Multon, A. Sellier, E. Grimal, F. Hamon, and E. Bourdarot, ‘Impact of stresses

and restraints on ASR expansion’, Constr. Build. Mater., vol. 140, pp. 58–74, Jun. 2017.

[11] R.-P. Martin, ‘Analyse sur structures modèles des effets mécaniques de la réaction sulfatique in-

terne du béton’, PhD thesis, Université Paris-Est, 2010.

[12] P. Chhun, A. Sellier, L. Lacarriere, S. Chataigner, and L. Gaillet, ‘Incremental modeling of re-

laxation of prestressing wires under variable loading and temperature’, Constr. Build. Mater.,

vol. 163, pp. 337–342, 2018.

[13] A. Sellier and S. Multon, ‘Chemical modelling of Delayed Ettringite Formation for assessment

of affected concrete structures’, Cem. Concr. Res., vol. 108, pp. 72–86, Jun. 2018.

effects of delayed ettringite formation on reinforced

Documents